In my prior article, I said ways to use Blue Databricks plus the Apache Spark collect_listing means to perform a-two-table relational data migration to NoSQL, by using the embedding approach to service a single-to-of numerous relationships. We put Apache Ignite because the during the time i did not have suitable local features in the Blue Investigation Warehouse (ADF) to support it conversion. Well, now we have it and is also (of course) titled assemble. This means will need several philosophy and you can aggregate him or her to your a keen array. We could explore collect in order to make arrays or a lot of time strings:
This post will reveal just how to migrate relational studies so you’re able to Azure Cosmos DB only using Blue Research Facility, with no password called for. Use situation is strictly like in 
my earlier in the day blog post, I am including they right here once more getting brief resource:
One-to-of several relationships with the embedding method
In a number of You to-to-Many conditions, the recommended approach is to try to Implant many side to the one front side, hence reducing the necessity for meets. A common analogy occurs when you will find a king/outline set of dining tables including Order Heading and you will Acquisition Outline.
Right here i’ve that listing into Purchase Header and you may about three associated information on Acquisition Detail. Inside the a beneficial relational business, we’re necessary to signup those two tables (by the SalesOrderID) to obtain an entire picture of conversion process study. When using the embedded method to move these records so you’re able to an enthusiastic Blue Cosmos DB (Key SQL API), the content will such as for instance just one file that have data having your order, and you will numerous points representing data into outline..
Note that I leftover the newest SalesOrderID function into embedded documents for site. The last implementation have a tendency to eradicate these types of issue because they are not required more.
The clear answer: migrating relational investigation
The answer provides just one Blue Research Factory pipeline with a great single Mapping Studies Flow passion you to checks out the latest relational research, transforms (embed) the data, last but not least lots the information and knowledge so you’re able to migrate relational data into Azure Cosmos DB. The final study flow need to look like this:
The fresh new DecimalToDouble sales is required as the Blue Cosmos DB cannot store Decimals with set accuracy. Which will make the desired Mapping Studies Move:
- Basic we add a few Research Supplies: Transformation Order Heading and you will Transformation Buy Detail. Optionally, we are able to set a great hash partition by the SalesOrderID into the both datasets on Enhance selection.
- After that, we add an Aggregate alter with the Conversion Buy Detail source group from the SalesOrderID. We’re going to add a single Aggregate line named Facts. This can is the articles we need to “embed”. Definitely tie the dwelling toward a pick-up means. The word on Details profession will be:
We have fun with toDouble here to ensure we do not upload decimals in order to Azure Cosmos DB. The data Preview to your the Aggregate action will want to look such as for instance this:
Implementation Notes
Playing with Azure Studies Facility Mapping Research Flows no-code method will make it really easy so you’re able to move relational investigation to Blue Cosmos DB. You need that it same way of create a lot more cutting-edge multi-level hierarchies otherwise carry out arrays off thinking when needed. Read more on how best to play with Assemble with Azure Cosmos DB.
Begin with Blue Cosmos DB
- Manage a separate membership playing with Azure Webpage, Sleeve template otherwise Blue CLI and relate solely to it with your favorite equipment.
- Remain up-to-big date for the current #AzureCosmosDB development and features following you to your Facebook The audience is really excited to see what you will create which have Blue Cosmos DB!
From the Blue Cosmos DB
Azure Cosmos DB are a completely addressed NoSQL databases to own progressive software development, having SLA-backed rate and you will supply, automatic and quick scalability, and you will open origin APIs to possess MongoDB, Cassandra, or other NoSQL motors.